Intel Bets on Cheaper AI as Foundry Losses Mount
This story is only covered by news sources that have yet to be evaluated by the independent media monitoring agencies we use to assess the quality and reliability of news outlets on our platform. Learn more here.19 Articles
19 Articles
Intel announces GPU D270, Xe3P architecture accelerator, focus on AI inference and up to 480 GB LPDDR5X memory
Intel has officially launched its new GPU for data centers and artificial intelligence acceleration, known by the codename "Crescent Island" and now presented as Intel GPU D270. Using the latest Xe3P graphics architecture, it can offer up to 480 GB of LPDDR5X. The accelerator board has exclusive focus on AI inference tasks, leaving aside resources dedicated to 3D graphics or games, even using an architecture capable of performing this type of pr…
We live in a time when AI has become the central point of interest of almost all companies and as such, it is moving a lot of money. Companies like NVIDIA have taken advantage of this moment to earn much more money by selling GPU for AI and prioritizing them before gaming. The same for AMD with its GPU Instinct and even Intel wants a part of the market. Intel was behind its two rivals, but its new GPU D270 Crescent Island is going to make looks …
Intel's new AI GPU uses the Xe3P architecture, features 480 GB of LPDDR5X memory, and a bandwidth of 1.5 TB/second.
Intel detailed its Crescent Island-based D-Series GPU accelerators with up to 1.3 PFLOPS computation theory FP8 and 1.5 TB/s bandwidth. The company points to AI’s inference on conventional air-cooled servers. Intel had planned customer sampling for the second half of 2026, but did not announce a commercial launch date.
Coverage Details
Bias Distribution
- No sources with tracked biases.
Factuality
To view factuality data please Upgrade to Premium
